Thermodynamics of molecular association. 9. An NMR study of

D. E. Martire, J. P. Sheridan, J. W. King, and S. E. O'Donnell. J. Am. Chem. Soc. , 1976, 98 (11), pp 3101–3106. DOI: 10.1021/ja00427a007. Publicati...
0 downloads 0 Views 772KB Size
3101 See footnote 15 in ref 8. W. J. A. Goossens, Mol. Crysf. Li9. Crysf., 12, 237 (1971). Reference 3b, Chapter IX. E. T. Samulski, Ph.D. Dissertation, Department of Chemistry, Princeton University, 1969. J. C. Powers and W. L. Peticolas, Biopofymers, 9, 195 (1970). S.Murthy, J. R. Knox, and E. T. Samulski, in preparation. M. Tsuboi, J. Polym. Sci., 59, 139 (1962). J. F. Yan, G. Vanderkooi, and H. A. Scheraga, J. Chem. Phys., 49, 2713 (1966). A. Tanaka and Y. Ishida, J. Polym. Sci.,Polym. Phys. Ed., 11, 1117 (1973). J. Hgarau, N. Lumbroso, and A. Pacault,. C. R. Hebd. Seances Acad. Sci., 242, 1702 (1956). C. M. French, Trans. faradaysoc., 50, 1320 (1954). P. W. Selwwd, “Magnetochemistry”, Interscience,New York, N.Y., 1956, Chapter VI. E. M. Bradbury, 8. G. Carpenter, C. Crane-Robinson, and H. Goldman, Netwe

(London), 225, 64 (1970). (38)M. Chien, E. T. Samulski, and C. G. Wade, Macromolecules, 6,638 (1973). (39)F. A. Bovey, “High Resolution NMR of Macromolecules”.Academic Press, New Ybrk, N.Y., 1972,Chapter XIII. (40)E. M. Bradbury, C. Crane-Robinson, L. Paolillo, and P. Temussi, Polymer, 14, 303 (1973). (41)M. Goodman, G. W. Davies, and E. Benedetti, Acc. Chem. Res., 1, 275 (1968). (42)A. A. Bothner-By and Gassend. Ann. N. Y. Acad. Sci., 222, 668 (1973). (43)Preliminary investigations of hyperfine interactions in anisotropic PBLG liquid crystals by 13C NMR indicate that segmental motions in the side chain are rapid (correlationtimes for reorientation < lO-’s). H. T. Edzes and E. T. Samulski, work in progress. (44)NOTE ADDED IN PROOF. Recent calculations of van der Waals-Lifshitz forces between chiral rods suspended in a dielectric medium indicate that PO-” for certain values of the delectric constant of the medium (E T Samulski, manuscript in preparation).

Thermodynamics of Molecular Association. 9. An NMR Study of Hydrogen Bonding of CHC13 and CHBr3 to Di-n-octyl Ether, Di-n-octyl Thioether, and Di-n-octylmethylamine D. E. Martire,* J. P. Sheridan, J. W. King, and S. E. O’Donnell Contribution from the Department of Chemistry, Georgetown Uniuersity. Washington, D.C. 20057. Received August 28, 1975

Abstract: Nuclear magnetic resonance studies in the temperature range 10 to 40 OC are reported for the hydrogen bonding between the electron acceptors and donors of the title, all in cyclohexane solution. Data on the variation of the chemical shifts of the haloform C-H proton as a function of donor concentration are treated by the Foster-Fyfe procedure which permits direct determination of the equilibrium constants, K , for 1 : 1 hydrogen-bonded complex formation. The results are discussed in terms of Deranleau’s criteria for reliability of the results and proof of fit of the proposed model. Enthalpies and entropies of hydrogen bonding are determined from the variation of In K with temperature. The thermodynamic constants so obtained and thermodynamic data previously determined by gas-liquid chromatography are used to evaluate thermodynamic parameters related to halogen/n-donor interaction. Consistent with other evidence and a polarization experiment reported here, the latter interactions are taken to involve negligible complex formation. The enthalpies of halogen/n-donor interaction are found to follow the trends CHBr3 > CHC13 and N > S > 0. The enthalpies of hydrogen bonding, which follow the trends CHCI3 > CHBr3 and N > 0 > S, are analyzed in terms of the double-scale equation of Drago.

In parts 2’ and 42 of this series we presented thermodynamic data (termed equilibrium constants, enthalpies, and entropies of “complex formation”) from gas-liquid chromatography (GLC) for the interaction of CHC13, CHBr3, and other haloalkane acceptors with four n-electron donors: din-octyl ether (DOE), di-n-octyl thioether (DOTE), di-noctylmethylamine (DOMA), and tri-n-hexylamine (THA). Analysis of the results obtained strongly indicated the presence of two types of interactions within these systems: (1) hydrogen bonding of the C-H hydrogen of the haloalkane to the n-donor atom, (2) charge transfer (n u* type) and/or electrostatic interaction between the n-donor and halogen atoms. It has been pointed out’ that thermodynamic measurements alone cannot serve to establish the existence of molecular complexes. Moreover, it has been shown recently3 that the generally accepted G L C methods for obtaining “association constants” yield, in fact, a quantity which is the sum of the equilibrium constant ( K , ) for all 1:l complex formation and a term ( a ) which reflects any noncomplexing acceptor-donor interactions present in the system, Le., “KGLC”= K t a. Hence, given the overwhelming spectroscopic evidence’ that haloalkanes (such as CHC13 and CHBr3) do indeed form hydrogen-bonded complexes with n-donors, we were left with two simple interpretations of the “KGLC”data: (1) “KGLC”= ( K K’), where K and K’ are the 1 : l association constants for

-

+

+

hydrogen-bonded and halogen-bonded complex formation, respectively, or (2) “KGLC”= ( K a ) ,where a is a measure of noncomplexing halogen/n-donor interaction or contact ~ a i r i n g .For ~ . ~lack of any conclusive evidence, one way or the other (see below), we arbitrarily chose the former description. Nevertheless, the qualitative discussion of the “KGLC” and concomitant enthalpy (which reflects a weighted average of all pairwise interactions) presented in parts 2 and 4 turns out to be equally valid for either description. The nature of halogen/n-donor interactions in these systems, however, is an unsettled Charge-transfer bands (attributed to n u* transitions) have been found in the ultraviolet region for several tetrahalomethane/amine s y ~ t e m s , ~ . ~ but the reliability of the derived K values ( K < 0.1 I. mol-’ for FCC13, ClCC13, and BrCC13 with trieth~lamine,~ and K x 0.03 1. mol-! for CC14/n-butylamine6) is open to Values this small indicate little or no complex formation and can hardly account for the “KGLC”values and trends observed in parts 2 and 4. Also, no charge-transfer bands have been reported1 or found in our laboratory for mixtures of CBr4 or CC14 and ether or thioether donors. Hence, while much thermodynamic, spectroscopic, and structural evidence’ exists for strong halogen/n-donor interactions, it is not clear whether substantial complex formation is involved, or whether charge-transfer interactions play an important role. With re-

+

-

Martire et al.

/

Thermodynamics of Molecular Association

3102 spect to the latter point, it should be emphasized that the mere appearance of a charge-transfer band is neither conclusive proof for the existence of complexes, nor proof that chargetransfer interactions (rather than electrostatic interactions) are predominant.’ Ideally, then, one would want to conduct studies that would permit partitioning of “KGLC”into its constituent parts, and perhaps shed some light on whether the aforementioned halogen interactions are better described in terms of complex formation or contact pairing. Of the variety of spectroscopic techniques employed in the study of hydrogen bonding, nuclear magnetic resonance (NMR) has proven to be one of the most sensitive and reliable, especially in the study of relatively weakly bonded systems such as those involving the C-H hydrogen. However, only a few of the studies that have been carried out have been quantitative from a thermodynamic point of view,’ most of them have been on chloroform 0nly,’3~and only one of them on any of the n-electron donors studied by GLC. Accordingly, the proton N M R technique is utilized here to obtain additional thermodynamic data on two of the haloforms previously studied by GLC, viz., chloroform and bromoform, with three n-electron donors: DOE, DOTE, and DOMA. Our expectation is that these data should relate directly to the hydrogen-bonding propensities of the haloforms to the various n-donors and hence (when combined with the G L C data) afford a method of estimating the contribution of halogenlndonor interactions. This expectation is based on our current view that the latter interactions probably involve negligible complex f ~ r m a t i o n . ~

Thermodynamic Association Parameters from NMR The equations presented below have been derived and discussed e l ~ e w h e r eThey . ~ are essentially based on an extension to NMR measurements of the Orgel-Mulliken4 treatment and interpretation of ultraviolet/visible measurements on supposed complexing systems. Consider a mixture of electron donor (D) and inert solvent (S) of concentration CD,to which is added a small amount of electron acceptor (A), where the condition CD >> CA obtains. Utilizing the molar concentration scale, an equilibrium or association constant ( K ) for the reaction A D 2 AD (1:l complex) may be written

+

(3) where A = (6 - &A), AAD = (&AD - 6 ~ ) AAD’ , = (&AD‘ - &A), and where K and K’ are the equilibrium constants for hydrogen and halogen bonding, respectively. (2) 6 is the weighted average of the proton chemical shifts corresponding to: (a) free and unperturbed A molecules, as in pure s(6~); (b) A molecules perturbed by 1 :1 hydrogen-bonded complex formation with D ( & A D ) ; (c) A molecules perturbed by noncomplexing interactions or contact pairing between the halogen atoms of A and D ( ~ A - D ) .The resulting equations are then:3 A/CD = - K A AC =

+ KAc

(4)

(KAAD+ ffaA-D)/K

(5) where A, AAD, and K are as defined above, AA-D = ( ~ A - D&A), and where the thermodynamic parameter CY (the nature of which is described elsewhere3) has been defined previously. Equations 2 and 4 are in the preferables Foster-Fyfe’ I form. Thus, for a series of solutions in which CDis varied, a plot of A/CD against A should be linear with ( K K’) (eq 2) or K (eq 4) being obtained directly as the negative gradient without recourse to an extrapolation. The value of Ac, which is concentration independent, may be obtained from the intercept with the ordinate which involves extrapolation to an infinitely dilute solution. The “KGLC” represent values corrected for minor donor nonideality.’,* Thus, if the model leading to eq 2 is applicable, one should find that “KGLC”= (K K‘), Le., that the negative gradient should be equal to the G L C “association constant”. Furthermore, if the temperature dependence of K is sufficiently different from that of K‘, then, according to eq 3, one should find AC to have a significant temperature d e p e n d e n ~ eeven ,~ if AAD’= 0. On the other hand, if eq 4 is applicable, the negative gradient should be less than “KGLc”, Le., “KGLC”> K and “KGLC”= (K a). Also, from eq 5, if the temperature dependence of K is sufficiently different from that of a,then AC should, in general, have a significant temperature dependence. However, for the case AA-D = 0, A, should be virtually temperature independent, assuming, of course, that AAD is relatively temperature independent.

+

+

+

Experimental Section

K=-- CAD Y A D CACDY A Y D

(1)

Materials. Reagent grade chloroform was washed with concentrated sulfuric acid, dilute sodium hydroxide, and water, until the washing was neutral. The chloroform was then dried over anhydrous where Ci and yi represent respectively the equilibrium conpotassium carbonate and distilled immediately before use (bp 61 .O centration and activity coefficient of component i, and where “C). Bromoform (bp 149.5 “ C ) was purified by the same procedure yi 1 as Ci 0. Given the dilute solution condition of A, and used for chloroform. The purification procedure for carbon tetralacking any other information, it is assumed that Y A D / Y A = bromide (used in a polarization experiment; see later) is described I . Also, while recognizing that, in general, it is often not valid elsewhere.’* Cyclohexane (the inert solvent) was Fisher Spectroanto neglect donor n ~ n i d e a l i t y , ~it. ’is~ assumed (and later jusalyzed reagent liquid and was used without further purification. The tified) that Y D = 1. Hence, the equilibrium constant is taken purification of the n-electron donors, di-n-octyl ether, di-n-octyl to be approximately equal to the equilibrium quotient CAD/ thioether, and di-n-octylmethylamine, is described elsewhere.1,2The concentrations of thepure donors at 30 “ C are: 3.299 (DOE). 3.238 CACD.Finally, it is assumed that all acceptor-donor reactions (DOTE), and 3.096 ( D O M A ) . all in mol I.-’ ( M ) . or interactions are rapid compared with the NMR time scale. Preparation of Samples. I n our experiments the haloform concenFor the systems considered here, we shall allow for two tration was kept constant a t 0.05 M, while the concentrations of the possible, simple interpretations of the observed proton chemical electron donors were kept in large excess and were varied from about shift of A (6) a t concentration CD. 0.5 to 2.1 M in five well-spaced concentrations. Solutions were made (1) 6 is the concentration-weighted average of the chemical up at room temperature by weighing the requisite amount of electron shifts corresponding to: (a) free A molecules, as in pure S ( 6 ~ ) ; donor in 10-ml volumetric flasks, adding 1 ml of a freshly prepared (b) A molecules perturbed by 1: 1 hydrogen-bonded complex haloform stock solution (0.5 M in cyclohexane), and enough cycloformation with D (&AD); (c) A molecules perturbed by 1 : l hexane to reach the I O ml mark. Samples were transferred to 5-mm halogen-bonded complex formation with D AD'). The re0.d. precision N M R tubes. For all samples, the tubes were filled to a height of at least 7 cm, sealed, and kept in the dark. The molarity of sulting equations are3 the solutions, at temperatures other than room temperature, was A calculated from the change in density with temperature of cyclohexane -= - ( K K’)A (K K’)Ac (2) and the respective electron donors.l%* CD

- -

+

+ +

Journal of the American Chemical Society

/

98:l I

/ May 26, 1976

3103

I

45

\

. \ \

I

->

60

80

A Figure 1. CHC13/DOMA: Linear plots of A/[D] vs. A at four temperatures, where [D] = CD. Negative of the slope equals K (see eq 4).

NMR Measurements. Proton N M R spectra were obtained a t four temperatures with a Bruker HFX-90 high-resolution spectrometer operating a t 90 M H z , in conjunction with a Bruker B-ST 100/700 variable temperature probe. Temperature measurements were carried out using a calibration curve involving the temperature dependence of the separation of the hydroxyl and methyl peaks of methanol.I3This method, which permits measurements to tenths of a degree, was employed several times during a set of runs. The temperature was observed to vary not more than f0.5OC, although it is possible that there may have existed a constant difference between the temperature observed with methanol and that existing in a given type of sample. Thermal equilibrium was attained by allowing I5 min before making measurements after insertion of a sample tube into the probe. Chemical shifts of the C - H signal in the haloform were measured by carrying out a frequency sweep with respect to cyclohexane used as an internal lock and a r e quoted as Hertz (Hz). The spectra were recorded at a sweep rate of 0.1 2 Hz/s. All signals are downfield from the cyclohexane reference. The precise frequencies were measured by means of a Hewlett-Packard H P 5216A 12.5 M H z electronic counter. Frequency sweeps were carried out in triplicate on all samples, and, in general, the results indicated that the precision of the measurements was about f O . l Hz.

Results In order to apply either eq 2 or 4 the concentration of one component of the complex must be small compared with the second component. In this respect a concentration of 0.05 M was chosen for both chloroform and bromoform, which is adequate to ensure a good signal-to-noise ratio while being low enough to satisfy eq 2 or 4 and also to ensure that no significant dimerization of the haloform O C C U ~InS the . ~ absence of donor, the chemical shifts of chloroform and bromoform (with respect to cyclohexane) were found to be constant a t 508.4 and 473.0 Hz, respectively, over the temperature range 10-40 OC. These frequencies were therefore taken to be 6 ~the , chemical shifts of the free and unperturbed monomer. Typical plots of A/CD against A for the complexes of DOMA with chloroform and bromoform are shown in Figures 1 and 2. Values for the negative gradient were determined by means of linear least-squares analysis and are summarized for all systems a t four temperatures in Table I. Since these values are clearly smaller than the corresponding ‘‘KGLc’’’*~ (see later), we shall commence to assume that they correspond to K values for 1 :1 hydrogen-bonded complex formation, Le., that eq 4 and 5 apply. The standard deviations in K range from 0.002 to 0.009, with a typical value being 0.005. Values of the enthalpy (AH) and entropy ( A S ) of hydrogen-bond formation

251

L--,20

60

40

~-

80

A

Figure 2. CHBr3/DOMA: Linear plots of A/[D] vs. A at four temperatures, where [D] = CD. Negative of the slope equals K (see eq 4). Table I. Equilibrium Constants K (I. mol-’) of Hydrogen Bonding System

10°C

21 OC

28OC

39OC

CHCI,/DOE CHBr3/DOE CHCI3/DOTE CHBr3/DOTE

0.398 0.324 0.358 0.381

0.321 0.279 0.321 0.326

0.285 0.253 0.285 0.309

0.226 0.212 0.252 0.282

CHC13/DOMA CHBr3/DOMA

0.478 0.523

0.371 0.412

29 “ C 0.301 0.339

0.242 0.282

Table 11. Enthalpies (kcal mol-’) and Entropies (cal mol-l der-’) of Hydrogen Bonding System

-AH

-AS

Ac, H z

CHC13/DOE CHBr3/DOE CHCI3/DOTE CHBr3/DOTE CHCI3/DOMA CHBr3/DOMA

3.40 f 0.09 2.56 f 0.10 2.17 f 0.12 1.80 f 0.12 4.16 f 0.07 3.77 f 0.07

13.8 f 0.3 11.2 f 0.3 9.7 f 0.4 8.3 6 0.4 16.1 f 0.2 14.6 f 0.2

83 f 2 81f1 77 f 1 65 k 1 153 f 2 136 f 2

-AH

System

-AH

System CHCI3In(C4H9)20“ CHCl,/tetrahydrofuranb CHBr3/tetrahydrofuranb CHC13/(C2H5)2S0

2.35 f 0.12 CHCI3/ (C2H5)2SC 3.6 f 0.4 CHC13/ (C2H5)3N0 2.6 f 0.2 CHCI3/ (C2H5)3Nd 1.70 f 0.02

2.2 f 0.3

4.05 f 0.03 4.2 f 0.2

Reference 9. C . J. Creswell and A. L. Allred, J . Am. Chem. SOC.,85, 1723 (1963). K. W . Jolley. L. M. Hughes, and 1. D. Watson, Aust. J . Chem.. 27, 287 (1974). C . J . Creswell and A. L. Allred, J . Phys. Chem., 66, 1469 ( 1 962).

were obtained by linear least-squares analysis of In K as a function of T-l according to the usual expression InK=-

-AH RT

+-ARS

The values of A H and A S and the corresponding standard deviations are listed in Table 11, along with enthalpy data determined by others for similar systems using N M R . With two Martire et al.

/

Thermodynamics of Molecular Association

3104 Table 111. Saturation Fraction System CHC13/DOE CHBr3/DOE CHC13/DOTE CHBr3/DOTE CHCI3/DOMA CHBr3/DOMA

IO

oc

0.17-0.46 0.17-0.41 0.15-0.43 0.16-0.45 0.18-0.49 0.19-0.52

(0.57)

(0.52) (0.54) (0.56) (0.60) (0.62)

39

oc

0.10-0.32 0.12-0.31 0.11-0.34 0.12-0.37 0.10-0.32 0.1 1-0.36

(0.42) (0.41) (0.45) (0.48) (0.43) (0.46)

See eq 7. Values in parentheses are the maximum attainable saturation fractions, i.e., pure donor values. (Pure donor concentrations are listed in ref 1 and 2.)

exceptions, the agreement is reasonably good. Included in Table I1 are the mean values of AC calculated at each temperature from the intercept with the ordinate of eq 4. N o systematic variation of AC with temperature was observed.

Discussion Deranleau8 has shown that equilibrium constants are most reliable when they are based on spectral data covering as much as possible of the saturation fraction ranges = 0.2-0.8, where (7) s being the saturation fraction of the dilute component (A).

Moreover, high uncertainty and little information result when s < 0.1. Summarized in Table 111 are the saturation fraction ranges a t the lowest and highest temperature studied, where the values in parentheses represent the maximum attainable saturation fractions, i.e., the values corresponding to the use of pure donor. While s 2 0.1 for all measurements, it is apparent that we did not and could not cover the recommended middle range. However, spot checks using 0.05 M solutions of CHC13 in pure donor revealed excellent accord with eq 4 and the results in Tables I ( K ) , I1 (Ac) and 111 (maximum s). The relatively small attainable saturation fraction range presents another potential problem apart from experimental error. Deranleau suggests8 that roughly 75% of the saturation fraction range be covered for adequate proof of fit of a given stoichiometric model, the primary concern here being the possibility of AD2 termolecular complexes, the substantial presence of which would produce curvature in the A / C D vs. A plot over a wide enough CD range. Although we have assumed that eq 4 rather than eq 2 applies, let us consider the possibility of AD2 complexes involving simultaneous hydrogen bonding and halogen bonding to the same acceptor. It appears improbable that, once a 1: 1 hydrogen-bonded complex has formed, a second donor molecule would be able to interact readily with the halogen atoms in view of the bulkiness of the alkyl side chains of the donors. Second, for the systems studied, over the concentration range studied, linear correlation coefficients in excess of 0.999 were found. Finally, in relation to the G L C experimental data, the question of AD2 complexes has been dealt with in considerable detaiL3>I4Suffice it to state here that current e v i d e n ~ e ’ ~(see . ’ ~ later) points to the unlikelihood of AD2 complexes for our six systems. Thus, we have carefully fit our N M R data over a reasonable concentration range to a specified model (i.e., allowing 1:l complexes only), which we regard to be the most realistic model. The validity of our assumption of donor ideality (i.e., YD 1 ) merits discussion. In the G L C experiment n-octadecane (OD) was used as the inert ~ o l v e n t . ’ Experimental ~~~’~ evidencel4.l5 in the form of linearity of solute (A) partition coefficient vs. C Dplots over a very wide concentration range (including pure D) fully supports our previous estimates’s2that YD 1 with O D as the solvent and DOE, DOTE, and DOMA as the electron donors. (It also supports our assumption that Journal of the American Chemical Society

the concentration of AD2 complexes is negligibly mall.^.^^) It might seem then, as it once did to us, that O D rather than cyclohexane should have been the inert solvent of choice for the N M R experiment. However, considerable line broadening (especially a t low C D )was found with O D (mp 28 “C), even for the strong singlet absorption of the haloforms. This may have been due to the relatively high viscosity of OD, compared to that of cyclohexane. Nevertheless, one set of experiments was conducted, with difficulty, on the system CHCI3/DOE/ O D a t 30 OC. A K value of 0.27 f 0.02 was obtained, i n excellent agreement with the cyclohexane value of 0.285 f 0.003 at 28 OC (Table 1) and an interpolated value of 0.269 a t 30 OC (Table IV), and in good agreement with 0.305 a t 30 OC, the value determined for the system CHC13/DOE/n-heptadecane.15 Thus, the evidence points to a rather small correction, if any, for donor nonideality when cyclohexane is used as the inert solvent .331° We have made and must examine the critical assumption that the N M R and G L C data are consistent with the model leading to eq 4 and 5, i.e., that “KCLC” = K c y , where K is the 1 : 1 equilibrium constant for hydrogen bonding and cy is a measure of noncomplexing halogen/n-donor interactions. Listed in Table IV are the values for K and a at I O OC intervals from 10 to 60 OC. They were calculated as follows. The ‘ ‘ K G L c ” ~(measured ,~ a t 30,40, 50, and 60 “C) were extrapolated to 10 and 20 O C by utilizing the linear least-squares fit of In KGLCas a function of T-I. Similarly, the K values from N M R were interpolated or extrapolated through eq 6, utilizing the least-squares values of AH and A S (Table 11). From the difference, cy was calculated. Also, the values listed in Table V were determined from -AH, AS ha=RT R where AHa may be regarded as a measure of the strength of halogen/n-donor interactions.’ I (For a discussion of the physical significance of a-related quantities and an example of their theoretical interpretation, the reader is referred else-

+

+*

here.^,^.' I )

W e note first that many of the cy values are substantial and that, at a given temperature and for a given donor, the a values for CHCI3 and CHBr3 are clearly different. Thus, consistent with the experimental evidence cited previously, a can hardly be attributed to the use of different inert solvents in the N M R and G L C experiments, i.e., to the nonideality of donor/cyclohexane mixtures in the N M R e ~ p e r i m e n t . ~In, ’ relation ~ to the observed temperature independence of Ac, given that the temperature dependencies of K (Table 11) and a (Table V ) are significantly different, we are lead by eq 5 to conclude that AA-D = 0, Le., AC = AAD. This implies that the proton supposedly “perturbed” by halogen/n-donor interaction is magnetically equivalent to a “free” haloform proton, a not unreasonable implication considering the separation of the former proton from the interacting centers. It should also be noted that eq 3 cannot account for the observed temperature independence of Ac. Additional corroborative evidence can be offered in defense of the aforementioned critical assumption that halogen/ndonor interactions result in negligible complex formation. First, spectrophotometric studies on tetrahalomethane/amine syst e m ~indicate ~ , ~ that, if complexes do indeed exist, the K values must be quite small ( CHC13 and D O M A > D O T E > DOE. This is significant and supportive of our model, in that thermodynamic, spectroscopic, and structural studies on haloalkane/n-donor systems] also indicate interaction strengths following the trends Br > CI and N > S > 0. An implication of our model is that the lifetime of halogen/n-donor interactions is of the order of the duration of a molecular collision, i.e., that the A-D “contact pair” has no extended lifetime as a separate entity.4 One might wonder, then, why it is that hydrogen-end collisions of the haloform with the n-donor lead to long-lived complexes, while halogen-end collisions apparently do not. This is particularly pertinent in view of the relative heats involved, the ranges being not terribly different: 1.8-4.2 kcal mol-’ for hydrogen bonding (Table 11) and 0.6-2.7 kcal mol-’ for halogen interaction (Table V). In interpreting dielectric absorption measurements on CC14/aromatic systems, North and Parkerlg suggested that it is the slope of the repulsive part of the pair interaction potential that governs the lifetime of the encounter rather than the “strength” ( A H ) of the interaction, viz., the greater the slope, the more “elastic” the collision, and the shorter the lifetime. They also suggested that charge-transfer interaction should lower the slope. Rationalizing our model in these terms then, the (angle-dependent) pair potential for halogen-end/ n-donor interaction should exhibit a steep repulsive slope, a not unreasonable possibility, considering the relative electronegativities of the halogen and n-donor atoms and the lack of any direct evidence for charge-transfer interaction with two of the

donors. To shed additianal light on the lifetimes of these collisions, dielectric absorption r n e a s ~ r e m e n t son ’ ~ A / D systems with A = CHC13, CHBr3, CC14, and CBr4 and D = ether, thioether, and tertiary amine would be of great use. Turning to the hydrogen bonding data and regarding A H as the only meaningful measure of hydrogen bonding “strength”,2o we see from Table I1 that the donor strength follows the sequence N > 0 > S, while the acceptor strength is always in the order CHC13 > CHBr3. These are the classical trends.] Thus, while it was difficult to discern definite trends in the weighted-average G L C enthalpies,Is2 the “separated” enthalpies ( A H and A H a ) show very clear and reasonable trends indeed. The above donor strength sequence was also found in a study of 18 aliphatic alcohol/n-donor systems,21 where the enthalpies were correlated with some success through the single-scale expression

I AHjjl

= Q;”Qjb

(9) where Qi? is the enthalpy parameter for acid i and Qjb is the enthalpy parameter for basej, both referred to an assigned Qb value of 1 .OO for DOTE. Qbvalues of 2.01 and 1.48 were found for DOMA and DOE, respectively.21 Analyzing the data in Table 11, we obtain acid parameters of 2.17 and 1.80 for CHC13 and CHBr3, respectively, and base parameters of 2.00 and 1.50 for DOMA and DOE, respectively. A more quantitative correlation of the enthalpy of hydrogen bonding is attainable through Drago’s double-scale equation21-23 -AH = E A E B

+ CACB

(10)

where E A and E B were interpreted as the susceptibility of the acid and base, respectively, to undergo electrostatic interactions, and CA and CBas the susceptibility to undergo covalent interactions. Using the recommended procedure23 and taking our previously estimated E B and CBvalues for DOE, DOTE, and DOMA2I as initial trial values, a set of acid and base parameters was found which minimized the differences between the retrieved (through eq 10) and experimental (Table I 1 here and Table V of ref 21) AH values. In all, 24 points were used to determine the 22 acid and base parameters listed in Table VI. With this set of parameters the retrieved and experimental values differed by an average of f0.07 kcal mol-’ for all systems and f 0 . 0 9 kcal mol-’ for the haloform systems, of the order of experimental error. The parameters are i n good agreement with those given for the bases di-n-butyl ether (CB = 3.40, E B = 1.06), diethyl thioether (CB= 7.40, E B = 0.34), and triethylamine (CB = 11.09, E B = 0.99), and the acid CHC13 (CA = 0.159, E A = 3.02).23It is particularly encouraging (and supportive of our model) that enthalpies of hydrogen bonding for aliphatic alcohol/n-donor systems from GLCZ1are compatible with N M R values for haloform/ndonor hydrogen bonding, and that both sets of data are compatible with the results of others. We see from Table VI that, in terms of its hydrogen-bonding characteristics, chloroform is quite similar to isopropyl alcohol and is a “harder” acid than bromoform. The sequence of effective base “hardness”, as measured by EB/cB,22’23 is D O E Martire et al.

/

Thermodynamics of Molecular Association

3106 Table VI. Acid and Base Parameters for Double-Scale Equation of Enthalpy of Hydrogen Bonding" CA

EA

0.183 0.156 0.234 0.259 0.184 0.150 0.159 0.155

2.96 2.59 2.72 2.72 2.58 2.39 2.54 1.95

tems.lV2 While the evidence a t hand indicates that halogen/ n-donor interactions do not result in significant complex formation, additional studies are needed to fully clarify the situation.

Acidsb n-Propyl alcohol Isopropyl alcohol n-Butyl alcohol Isobutyl alcohol sec-Butyl alcohol tert-Butyl alcohol C h loroform Bromoform Basesb DOE

DOTE DOMA a

3.01 7.71 11.50

1.11 0.34 0.97

See eq 10. Alcohol and base values a r e revised values; see ref

21.

> D O M A > DOTE. Taken a t face value, the results indicate that, for the two haloforms, electrostatic forces account for roughly 85% of the total hydrogen bonding energy with DOE, about 55% with DOMA, and about 40% with DOTE, in accordance with the above sequence. Also, although the absolute electrostatic contribution ( E A E B to ) - A H is greater for the DOE systems, the net -AH is greater for the DOMA systems, due to the substantial covalent parameter of the amine. In summary, the simple model leading to eq 4 and 5, the N M R results for hydrogen bonding, auxiliary experimental information, and some useful concept^^^^-'^^** have permitted us to further resolve and understand thermodynamic data previously determined by GLC for haloform/n-donor sys-

Acknowledgment. This research was supported by a grant from the National Science Foundation. Helpful discussions with E. D. Becker are gratefully acknowledged. G . M. Janini is thanked for his assistance with the polarization measurements. References and Notes (1) J. P. Sheridan, D. E. Martire, and Y. B. Tewari, J. Am. Chem. Soc..94, 3294 (1972), and references therein. (2) J. P. Sheridan, D. E. Martire, and F. B. Banda, J. Am. Chem. SOC.,95,4788 (1973). (3) D. E. Martire, Anal. Chem.. 46, 1712 (1974). (4) L. E. Orgel and R. S.Mulliken, J. Am. Chem. SOC.,79, 4839 (1957). (5) D. P. Stevenson and G. M. Coppinger, J. Am. Chem. SOC.,84, 149 (1962). (6) C. J. Biaseile and J. G. Miller, J. Am. Chem. SOC.,96, 3813 (1974). (7) W. B. Person, J. Am. Chem. SOC.,87, 167 (1965). (8) D. A. Deranleau, J. Am. Chem. SOC.,91, 4044, 4050 (1969). (9) G. R . Willey and S. I. Miller, J. Am. Chem. SOC.,94,3287 (1972). (10) M. W. Hannaand D. G. Rose, J. Am. Chem. Soc., 94,2601 (1972). (11) R. Foster and C. A. Fyfe, Trans. faraday SOC.,61, 1626 (1965). (12) Part 8 in this series: G. M. Janini. J. W. King, and D. E.Martire, J. Am. Chsm. Soc., 96, 5368 (1974). (13) A. L. Van Geet, Anal. Chem., 40, 2227 (1968). (14) H. L. Liao, D. E. Martire, and J. P. Sheridan, Anal. Chem., 45, 2087 (1973). (15) J. H. Purnell and J. M. Vargas de Andrade, J. Am. Chem. Soc., 97, 3585, 3590 (1975). (16) E. A. Guggenheim, Trans. FaradaySoc., 45, 714 (1949). (17) J. W. Smith, Trans. FaradaySoc., 46, 394(1950). (18) R. K. Chan and S.C. Liao, Can. J. Chem., 48, 299 (1970). (19) A. M. North and T. G. Parker, Trans. Faraday Soc., 67, 2234 (1971). (20) M. L. McGlashan, D. Stubley, and H. Watts, J. Chem. SOC.A, 673 (1969). (21) H. L. Liao and D. E. Martire, J. Am. Chem. Soc., 96, 2058 (1974). (22) R. S.Drago. G. C. Vogel, and T. E. Needham, J. Am. Chem. Soc., 93,6014 (1971). (23) R. S. Drago. "Structure and Bonding", Vol. 15, J. D. Dunitz et al, Ed., Springer-Verlag, Berlin, 1973, p 73.

Coupling Constants in Carbocations. 11. Angular Dependence of ~ J CinH Groups Adjacent to Cationic Carbons. A New Criterion for Interpreting NMR Spectra of Carbocations 13C-lH

David P. Kelly,** Graham R. Underwood, and Peter F. Barron Contribution from t h e D e p a r t m e n t of Organic Chemistry, Unioersity of Melbourne, Parkcille, Victoria 3052, A u s t r a l i a . Receired Julj' 27, I975

Abstract: The values of ' J C Hin groups adjacent to cationic carbons in classical, static, carbenium ions a r e dependent upon the dihedral angle between the C-H bond and the unoccupied p orbital of the cationic carbon. This dependence is given by AJ = A - B cos2 6 , where AJ is the difference in Jctj between that in the cation and that in a neutral model compound (ketone), A is the maximum inductive enhancement of J C H (22.5 H z ) , and B is the maximum hyperconjugative diminution of J C H (33.1 H z ) . The equation provides another criterion by which chemical shift assignments may be made. In the case of the 2-methylnorbornyl cation, some discrepancies in the previous assignments have been corrected. Comparison of the shift of the methyl carbon of this cation with that in the corresponding alcohol shows that this carbon experiences no deshielding upon ion formation. The significance of this finding is discussed. The equation is applicable to some equilibrating classical cations, and can be used to determine conformations in cations, for example, the rert-amyl cation.

In our investigation of the applicability of JCHa s a criterion for the presence of a bridging in cyclopropylcarbinyl cations,l it was necessary to determine the effect of the adjacent positive charge on the apical, methine coupling constant in classical (non-a bridged) cyclopropylcarbinyl cations. Thus the J C H values increase in the tertiary cations 1 and z3 from those in

'

Journal of the American Chemical S o c i e t y

98:ll

the neutral model compounds 4 and 5 by 25 and 22 Hz, respectively ( A J ) . The ketones are considered to be appropriate model compounds since they contain a trigonal carbon and are therefore more appropriate models than the corresponding hydrocarbons or carbinols.' A similar, large increase occurs for J C ! Hin the 2-methylnorbornyl cation 34 from that in 2-

1 May 26, I976